Add Enhanced DLP Analytics and Cross-Workload Correlation Plugin

[AKT_18 (arunt14)]

Summary

Adds a new KQL-based plugin with 7 advanced DLP analytics skills for cross-workload correlation, false positive analysis, policy health monitoring, and sensitivity label compliance tracking. Also adds 12 new sample prompts and comprehensive documentation.

Intended Audience

• SOC Analysts investigating multi-workload DLP incidents
• Data Security Administrators tuning DLP policy effectiveness
• Compliance Officers assessing policy coverage and label compliance
• Security Managers monitoring organizational DLP posture

What This Adds

Plugin: DLPEnhancedAnalytics

Location: Plugins/Community Based Plugins/Purview/KQL_DLPEnhanced.yml
Documentation: Plugins/Community Based Plugins/Purview/KQL_DLPEnhanced_README.md

#	Skill	Purpose
1	DLPCrossWorkloadCorrelation	Unified view of user DLP violations across Exchange, Teams, SharePoint, OneDrive, Endpoints
2	DLPFalsePositiveAnalysis	Override rate analysis with risk classification (High/Medium/Low/Minimal)
3	DLPPolicyCoverageGaps	Detect workloads with zero DLP coverage — flags critical gaps
4	DLPHighRiskUsersAcrossWorkloads	Multi-workload violation flagging with risk scores (Critical/High/Medium/Low)
5	DLPAlertTrendAnalysis	Daily time-series trends with severity distribution for spike detection
6	DLPLabelDowngradeRemoval	Track sensitivity label downgrades/removals (DLP bypass detection)
7	DLPPolicyHealthCheck	Policy health dashboard: match rates, override rates, Healthy/Unhealthy status

Updated Sample Prompts

Location: Sample Prompts/Microsoft Purview/Readme.md

Added 4 new categories with 12 sample prompts:

• Cross-Workload DLP Correlation — correlate alerts across all platforms
• DLP False Positive Analysis — analyze override rates and policy effectiveness
• DLP Policy Health and Coverage — health checks and coverage gap detection
• Sensitivity Label Compliance — label downgrade and removal tracking

Prerequisites

• Security Copilot Enabled
• Access to upload custom plugins
• Microsoft Defender XDR with CloudAppEvents data ingestion enabled
• DLP policies configured across workloads
• Sensitivity labels configured for label tracking skills

Supported Microsoft Products

• Microsoft Security Copilot
• Microsoft Purview Data Loss Prevention
• Microsoft Defender XDR

Relationship to Existing Plugin

This plugin complements the existing KQL_DataAnalyst.yml without overlap:

Capability	Existing Plugin	This Plugin
Per-workload DLP alerts	✅ 15 skills	—
Sensitivity file monitoring	✅ 16 skills	—
DLP policy tuning	✅ 5 skills	—
Cross-workload correlation	❌	✅
False positive analysis	❌	✅
Policy coverage gaps	❌	✅
High-risk user scoring	❌	✅
Alert trend analysis	❌	✅
Label downgrade tracking	❌	✅
Policy health dashboard	❌	✅

Sample Prompts

Correlate all DLP alerts for user user@contoso.com across all workloads in the last 14 days
Which DLP policies have the highest false positive override rates?
Identify DLP policy coverage gaps — which workloads have no protection?
Show me users with DLP violations across 2+ workloads — who are highest risk?
Show me DLP alert trends over the last 30 days — are there any spikes?
Which users have downgraded or removed sensitivity labels recently?
Run a DLP policy health check — show effectiveness for all policies

Limitations

• Results limited to CloudAppEvents data retention period in your tenant
• Cross-workload correlation requires DLP policies enabled on the relevant workloads
• False positive analysis uses override rates as a proxy — high overrides may also indicate user training gaps
• Policy coverage gap detection identifies workloads with zero matches; no matches could also mean no sensitive data activity
• Read-only analytics — does not modify policies, labels, or permissions

Responsible AI

• Plugin provides read-only security analytics and investigation capabilities
• No automated remediation, policy changes, or user restrictions
• Risk scores and health statuses are computed metrics requiring analyst interpretation
• Compliant with Microsoft Responsible AI Publishing Requirements